Arbor is a free MCP server tool developed by Arbor Technologies Limited for the mcp platform, categorized under mcps. Built primarily with Rust, it serves as a structural intelligence layer for AI agents and LLMs, providing graph-native code intelligence. It connects to codebases, helping both human developers and AI agents navigate, analyze, and perform complex operations within software projects. Arbor is distributed via Homebrew for macOS, by building from source, or through release bundles containing its core binaries.

What does Arbor connect to?

The Arbor MCP server connects to AI agents and LLMs, offering a structural intelligence layer and graph-native code intelligence for codebases. It operates as part of a shared daemon, which also powers a desktop application, web UI, and CLI components. This architecture enables it to integrate deeply with various development environments and intelligent systems. It supports stdio as the default transport protocol and HTTP (including Streamable HTTP and Server-Sent Events or SSE) for remote or enterprise deployments.

Core features and tools

Arbor exposes a range of callable tools for interacting with codebases. These include Orientation with get_map; Surgical tools like list_entry_points, get_callers, get_callees, search_symbols, get_file_graph, and get_node_detail; and Broad capabilities such as get_logic_path, analyze_impact, find_path, and get_knowledge_path. Additionally, Agent-native tools are provided, including get_blast_radius, explain_symbol, audit_security, get_architecture_overview, and batch_query.

Who is it for?

Arbor is designed for human developers and AI agents, including coding agents and LLMs, engaged in software development. Its capabilities assist with tasks requiring deep code understanding, dependency tracing, and impact analysis. Compatible AI clients include Claude Code, Cursor, VS Code, Codex, and OpenCode. It suits environments where AI agents need to interact with codebases intelligently. This is not the right fit if your workflow does not involve AI agents or LLMs needing structural or graph-native code intelligence.

How to get started

The server authenticates using ARBOR_DAEMON_AUTH_TOKEN for remote authenticated daemons, typically employing a bearer token. For remote HTTP/HTTPS connections, access is granted through a standard OAuth 2.1 authorization flow with Proof Key for Code Exchange (PKCE).

Use cases

• Issue-driven coding: Creating and managing worktrees from issues, running embedded terminals, and inspecting pull request context.
• Code navigation and analysis: Allowing AI agents to locate code, trace dependencies, and analyze the impact of changes within a codebase.
• Refactoring and security auditing: Enabling AI agents to perform complex refactors, assess the blast radius of changes, and conduct security audits.

Arbor provides a critical structural intelligence layer for both human developers and AI agents working with codebases. Its array of surgical, broad, and agent-native tools supports complex software development workflows, from impact analysis to security auditing. The ability to connect to various AI clients and transport protocols makes it adaptable for diverse deployments. Try analyzing the impact of changes using its analyze_impact tool to understand its capabilities.